Emotional regulation neural circuitry abnormalities in adult bipolar disorder: dissociating effects of long-term depression history from relationships with present symptoms

Bipolar disorder (BD) is common and debilitating and confounding effects of depression history on neural activity in BD are unknown. We aimed to dissociate neural activity reflecting past depression-load vs. present symptom severity using the Course and Outcome of Bipolar Youth (COBY), a prospective longitudinal cohort study of pediatric-onset BD. In n = 54 COBY (18–32 years), we modeled depression scores over time (up to 17.5 years) using a standardized autoregressive moving average (ARMA) model, followed by k-means cluster analysis. N = 36 healthy participants (HC, 20–36 years) were included. Using two factorial analyses, we parsed the impact of ARMA-defined past depression-load on neural activity from the impact of current symptoms on neural activity (p < 0.001, k > 30) and examined relationships with past and present symptoms (ps FDR-corrected). ARMA identified three COBY groups based on past depression-load. ARMA-defined COBY participants with the greatest past depression-load vs. other groups showed greater activity in right temporoparietal junction, thalamus, insula, premotor cortex, left fusiform gyrus, bilateral precuneus and cerebellum. In contrast, BD-COBY participants vs. HC showed greater activity in left hippocampus, dorsolateral prefrontal cortex, and right somatosensory cortex, plus the above thalamus, premotor cortex and cerebellum; activity positively correlated with present symptom severity in most regions. Past depression-load was related to social cognition and salience perception network activity, potentially reflecting heightened attention to socially relevant distracters, while present symptoms were associated with emotion processing and reappraisal network activity, potentially reflecting abnormal emotional experience and regulation. Differentiating aberrant neural activity related to long-term depression vs. present affective symptoms can help target interventions to networks associated with pathophysiological processes, rather than long-term illness effects.